Organisations have been automating aspects of their businesses and operations for more than 250 years. All of that historical effort, though, has been spent on building automated systems that then require us to re-shape our behaviours, our teams, our work environments and our processes around them. Our 250 years’ worth of automated systems have shaped every aspect of our businesses, and not necessarily in a good way. And this is as much the case in an office environment as it is on a production line.
However, a new wave of automation is now breaking on the shores of office environments across industries — from financial services to manufacturing and utilities, and everywhere in between. Specifically, what’s happening is that our historical, statically designed and built automations are being augmented by artificial intelligence (AI) technologies that can bridge the gap between the predictable, binary digital world (where automations naturally exist and perform well) and the messy, unpredictable, analogue human world. This is “intelligent automation”.
Intelligent automation increases the potential for automated systems to sense, understand and act on information from the physical world, as well as to learn and adapt — and thereby address more types of work tasks.
As an example: In a back-office business environment, an automated system can pull invoice data from one system or data source, carry out some automated checks on the validity of each invoice, attempt to match it against a purchase order, and then update the accounting system of record. With an intelligent automation approach, that core functionality can be augmented to increase the scope and value of the automation. The system can work through a collection of scanned paper documents as well as PDFs, XML-based system-readable documents and more; use a variety of computer vision, optical character recognition (OCR), document classification and object extraction techniques to identify which documents are true invoices; extract the important information needed from each invoice regardless of format (supplier details, line items, totals, payment terms and so on); carry out advanced checks against not only purchase orders but also broader commercial contracts and other agreements; and, where it cannot automate these activities, learn from the input of a human expert to tweak its algorithms.
One insurance company employed an intelligent automation strategy to help it reduce the manual processing of inbound email communications from its network of brokers. It’s saved 2,000 person-hours per month, managing to automate 98 per cent of the email processing work and reducing the cost per transaction by 91 per cent. By freeing up staff to work on more fulfilling tasks, staff satisfaction has also improved.
This example is of an automated system that works by itself, behind the scenes — and there are huge opportunities to streamline and improve the quality of work in this kind of configuration across finance, human resources, procurement, operations and other support functions. However, this kind of “unattended” automation configuration is not the only way in which intelligent automation will be delivered. It’s also clear that there are huge opportunities to use intelligent automation to create intelligent digital co-workers that act alongside humans in a knowledge-work environment. Indeed, by 2024, IDC predicts that 50 per cent of structured repeatable tasks will be automated, and 20 per cent of workers in knowledge-intensive tasks will have AI-infused software or some other digitally connected technology as a co-worker.
Of course, this new wave of automation will also have an impact on working practices. The opportunities to improve business productivity, accuracy, quality, business intelligence/decision making and agility are profound — but organisations cannot overestimate the importance of carefully thinking through the ways in which intelligent automated systems and people will need to be shaped and empowered, respectively, to adjust their activities and behaviours around each other.
Beyond the criticality of properly considering data quality and availability when designing and training intelligent automations, a human-centric approach to designing and introducing these new intelligent automations will be critical — and essential if we are to avoid playing into lazy stereotypes of “humans against machines“.